Benedikt Perak, Tajana Ban Kirigin

Paper presentation at: Logic and Applications (LAP)

LAP 2020 was held as a hybrid meeting, online and in Dubrovnik, September 21 – 25, 2020.


The paper will demonstrate the ConGraCNet application (ConGraCNet, for distinguishing word senses and identifying semantically related lexemes in a corpus by using the syntactic-semantic patterns of language usage. This unsupervised tagged corpus graph analysis method is based on the construction grammar approach to syntactic dependencies. ConGraCNet relies explicitly on the coordinated [x and|or y] (Van Oirsouw 2019, Dorow and Widdows 2003) and [x_is_a_y] syntactic grammatical relations between the lexical co-occurrences for the construction of the network representation. For a given source lexeme in a corpus, the method yields associated communities of collocation lexemes that represent the sense structure and different meanings based on the context of its usage. By projecting semantic value to a coordinated syntactical relation [x and|or y], we can filter the lexical collocates with high conceptual similarity from a corpus and construct clustered lexical networks that reveal ambiguous referential meanings of a source lexeme. The members of a cluster are processed with and iterative graph function that finds best candidates for abstracted class label using [x_is_a_y] syntactic-semantic construction. 

For instance, the lexeme assertiveness-n with 20809 occurrences in English Timestamped JSI web corpus 2014-2019, when processed with n=15 collocates used to construct a second-degree coordination graph, pruned with: degree >= 2, clustering method: leiden, partition type: mvp yields network of 43 elements and 4 clusters (Figure 1). The [x_is_a_y] syntactic-semantic construction reveals the class labels in first and second degree. In relation with the members of a class 1: [‘self-confidence-n’ ‘self-esteem-n’ ‘confidence-n’ ‘self-advocacy-n’ ‘self-worth-n’ ‘self-respect-n’ ‘self-image-n’ ‘esteem-n’ ‘self-awareness-n’ ‘self-reliance-n’ ‘independence-n’ ‘pride-n’ ‘Pride-n’ ‘motivation-n’ ‘skill-n’], ASSERTIVENESS-N is related to [‘PRIDE-N’, ‘PRIDE-N’]1 [‘EMOTION-N’, ‘MOTIVATION-N’]2. In relation with the members of a class 2: [‘aggressiveness-n’ ‘aggression-n’ ‘sociability-n’ ‘talkativeness-n’ ‘impulsivity-n’ ‘passivity-n’ ‘hyperactivity-n’ ‘hostility-n’ ‘irritability-n’ ‘impulsiveness-n’ ‘shyness-n’ ‘restlessness-n’ ‘agitation-n’ ‘euphoria-n’], ASSERTIVENESS-N is related to [‘AGGRESSION-N’, ‘TRAIT-N’], [‘FEAR-N’, ‘REACTION-N’]2. In relation with the members of a class 3: [‘decisiveness-n’ ‘directness-n’ ‘boldness-n’ ‘optimism-n’ ‘extraversion-n’ ‘courage-n’ ‘bravery-n’ ‘clarity-n’ ‘frankness-n’], ASSERTIVENESS-N is [‘COURAGE-N’, ‘BRAVERY-N’] 1, [FAITH-N’, ‘VIRTUE-N’]2 In relation with the members of a class 4: [‘assertiveness-n’ ‘cooperativeness-n’ ‘dominance-n’ ‘listening-n’ ‘friendliness-n’], ASSERTIVENESS-N is related to [‘LISTENING-N’, ‘SKILL-N’] 1, [‘ISSUE-N’, ‘PROBLEM-N’]2.

We will explain the impact of the modulation of the linguistic and graph parameters, exemplify the application of the procedure on several lexemes in different languages and corpora and present the implementation of the  WordNet external knowledge databases for further refinement of the results.


ConGraCNet application. (2020).

Sketch Engine.

Van Oirsouw R.R. (2019) The syntax of coordination. Routledge

Dorow B.  and Widdows D.(2003).  Discovering corpus-specific word senses, 10th Conference of the  European Chapter of the Association for Computational Linguistics

Link to the presentation